An extensive Markov process for ECG noiseless coding

In the paper, an extensive Markov process for noiseless ECG data compression is presented. It is an extension of the Markov process to measure the entropy value of a discrete signal. The extensive Markov process utilizes the inter sample correlations by predicting the incoming three samples based on the previous three samples which constitute an extensive Markov process state. In every state, there is a Huffman code table for the third extension code of the incoming three input samples, and the sizes of-the tables vary from state to state. Results show that an average of compression ratio (Cr) being 4.550 can be achieved. This corresponds to an average of 2.6376 bits/sample for each ECG signal, while the average entropy value for the original signals is 8.396 bits/sample. Although many noiseless ECG data compressors have been presented before, few has its Cr being greater than 4.0.<>